I just loaded a 15000x7500 pixel panoramic image (or environment map) into Hitfilm 3 Pro - created with Microsoft ICE- dragged the Environment Map Effect onto it, messed about with the scale etc. and made a very pleasant spinning view with a couple of keyframes on the rotation.
So far, so good, it plays at close to realtime speeds and is very smooth, possibly dropping one frame in 30 to keep the speed up.
As there was recently a question about the size of assets you could load into Hitfilm 4 Express, I loaded the same project into that.
It worked, but...motion was a lot jerkier on playback, it looks like it's dropping one frame in 10 or 12. Even with a RAM preview, it was still slightly jerky, which rather defeats the purpose of the preview.
Back to HF3P again and I RAM previewed that and it was not only faster to make it (felt like 1.5x-2x faster, certainly a very noticeable speed improvement) play back was buttery smooth.
So back to HF4E and...same jerkiness again.
How about Hitfilm 4 Pro?
Holy moley! It's just awful. It looks like it's dropping about 3 out of every 4 frames and is like watching a jerky stop motion animation.
RAM Preview? As slow to create as HF4E and plays back with the same occasional jerkiness and juddering.
I know HF4P has lots of extra 'whizzybangs', but at every single stage the performance is slower than HF3P and it's disappointing to find yet another example (proxy creation is another).
It's surprising though that as they share a common(ish) engine, that HF4E is faster than HF4P. Maybe no 3D models handling in the codebase is actually an advantage?
I couldn't test it in HF3E as the Environment Map Effect isn't available in that.
I flipped back and forth between all the 3 versions that would load the project and the speed differences were consistent. Motion Blur was turned off in the Composite, BTW.
As @NormanPCN said to @JoshDaviesCEO in this thread here:
RAM Preview "You are painting far too broad a brush stroke by saying that GPU read back is "poor at best"."
Whatever the issue is, it doesn't seem to apply to HF3P, so what other reason might there be for HF4E, and especially HF4P's, overall poor performance?